Ph.D. thesis (pdf) - dirac

173
temperature in the harmonic approximation, but it differs by mean square displacement
in two respects. There is no average over Q and it is a relative quantity which
compares the amplitude of vibrations to the frozen in density fluctuations. The
latter does not change much as a function of temperature in the glass. However, the
frozen in fluctuations will also decrease as pressure is increased. This means that
two competing effects govern the pressure dependence of the nonergodicity factor.
In practice we find that the nonergodicity factor measured in the glass is essentially
independent of pressure at a given temperature. Thus the data at different pressures
essentially collapse (at least at low T) when comparing the temperature dependence
on an absolute temperature scale. Scaling the temperature axis with the pressure
dependent glass transition temperature therefore makes the curves separate and it
has the consequence that the dimensionless parameter α increases when pressure
increases. This increase of α is opposite the behavior expectation from the correlation
between α and isobaric fragility. Also our data on different molecular weight
polymers is in contradiction with the proposed correlation between α and fragility.
As an alternative we suggest based on a combination of our data and literature
data that the original finding is a consequence of a “hidden” correlation between
the nonergodicity factor and the effect of density on the relaxation time. We find
that f(T g ) is smaller when the effect of density on the relaxation time is larger.
This means that the vibrational part of the density fluctuations in the considered
Q-range are larger when the effect of density on the relaxation time is larger. This
suggests that the properties which govern these density fluctuations also couple to
the density dependence of the relaxation time. More studies are needed to verify this
interpretation - particularly studies of the pressure dependence of f(T g ) are needed.
Following the work of Inamura et al. [2000, 2001] it has been thought that the boson
peak intensity decreases when a sample is densificated. However, this work and
several other results that have come out within the last year [Monaco et al., 2006 a;
Andrikopoulos et al., 2006], clearly demonstrate that this is not the case. The boson
peak intensity relative to the Debye density of states stays constant (or increases)
as pressure is increased. The system becomes harder overall, but no specific modes
appear to be suppressed. Also the shape of the boson peak stays constant as pressure
is increased. In so far as the boson peak gives a signature of the type of disorder, this
suggests that the type of disorder is not changed when the system is compressed.
Both the boson peak intensity and the quasi-elastic scattering measured directly in
the dynamical structure factor at T g decrease as a function of pressure. Moreover,
both contributions change by the same ratio along the T g -isochrone and the relative
intensity of the boson peak as compared to the quasi-elastic intensity therefore stays